Quasar Clustering at $25\kpch$ from a Complete Sample of Binaries
Adam D. Myers, Gordon T. Richards, Robert J. Brunner, Donald P. Schneider, Natalie E. Strand, Patrick B. Hall, Jeffrey A. Blomquist, Donald G. York
aa r X i v : . [ a s t r o - ph ] S e p ApJ, in prep November 11, 2018
Preprint typeset using L A TEX style emulateapj v. 08/22/09
QUASAR CLUSTERING AT 25 h − kpc FROM A COMPLETE SAMPLE OF BINARIES Adam D. Myers , Gordon T. Richards , Robert J. Brunner , Donald P. Schneider , Natalie E. Strand ,Patrick B. Hall , Jeffrey A. Blomquist , and Donald G. York ApJ, in prep November 11, 2018
ABSTRACTWe present spectroscopy of binary quasar candidates selected from Data Release 4 of the SloanDigital Sky Survey (SDSS DR4) using Kernel Density Estimation (KDE). We present 27 new sets ofobservations, 10 of which are binary quasars, roughly doubling the number of known g <
21 binarieswith component separations of 3 ′′ ≤ ∆ θ < ′′ . Only 3 of 49 spectroscopically identified objects arenon-quasars, confirming that the quasar selection efficiency of the KDE technique is ∼ u − g < . < R prop < . h − kpc, we determine the projectedquasar correlation function to be W p = 24 . ± . . , which is 2 σ lower than recent estimates. We arguethat our low W p estimates may indicate redshift evolution in the quasar correlation function from z ∼ . z ∼ . R prop ∼ h − kpc. The size of this evolution broadly tracksquasar clustering on larger scales, consistent with merger-driven models of quasar origin. Althoughour sample alone is insufficient to detect evolution in quasar clustering on small scales, an i -selectedDR6 KDE quasar catalog, which will contain several hundred z ∼ < R prop ∼ h − kpc. Subject headings: cosmology: observations — large-scale structure of universe — quasars: general —surveys INTRODUCTION
Cosmologically, quasars can now be explained as onespectacular stage of an evolutionary process, initiated bygas-rich galaxy mergers, that ultimately helps redden el-liptical galaxies (see, e.g., Hopkins et al. 2006, 2007a).That quasar activity might trace the early stages ofmerger-driven galaxy evolution makes quasar observa-tions an essential ingredient in constraining galaxy for-mation scenarios. On the other hand, less luminous Ac-tive Galactic Nuclei (AGN), particularly at low redshift( z ∼ < Electronic address: [email protected] Some data presented here were obtained at Kitt Peak NationalObservatory, a division of the National Optical Astronomy Obser-vatories, which is operated by the Association of Universities forResearch in Astronomy, Inc. under cooperative agreement withthe National Science Foundation. Department of Astronomy, University of Illinois at Urbana-Champaign,Urbana, IL 61801 Department of Physics, Drexel University, 3141 ChestnutStreet, Philadelphia, PA 19104 National Center for Supercomputing Applications, Cham-paign, IL 61820 Department of Astronomy and Astrophysics, 525 Davey Labo-ratory, Pennsylvania State University, University Park, PA 16802 Department of Physics, University of Illinois at Urbana-Champaign, Urbana, IL 61801 Department of Physics and Astronomy, York University,Toronto, ON M3J 1P3, Canada Department of Astronomy and Astrophysics, University ofChicago, Chicago, IL 60637 that orientation can complementarily explain many dif-ferences between the AGN zoo (e.g., Antonucci 1993;Elvis 2000), new AGN constraints over a broad range ofluminosity, particularly across z ∼
1, may be key to de-termining which elements of quasar behavior are mainlystructural, and which are mainly evolutionary.If quasars are associated with galaxy mergers, obser-vations of binary quasars with proper separations com-parable to the scale of small galaxy groups should of-fer interesting constraints. It has generally become ac-cepted that most quasar pairs at similar redshifts thathave image separations ∼ > ′′ , are binary quasars ratherthan lenses (Phinney & Blandford 1986; Bahcall et al.1986; Kochanek et al. 1999; Rusin 2002; Oguri 2006).AGN activity can be exacerbated by tidal forces ingalaxy mergers (Barnes & Hernquist 1996; Bahcall et al.1997) and it has long been argued that this mightexplain an excess of binary quasars (Djorgovski 1991;Kochanek et al. 1999; Mortlock et al. 1999) and physicaltriples (Djorgovski et al. 2007). Hopkins et al. (2007b)suggest instead that a binary quasar excess simply re-flects the increased probability for mergers to occur inregions that are overdense on small scales. If quasarsform in mergers they will thus be naturally more biasedat small scales. Hopkins et al. (2007b) further argue thatorbits for which quasar activity might be exacerbated inboth of two merging galaxies are prohibitively rare, evenif few such events are needed to explain a binary quasarexcess (Myers et al. 2007b; henceforth M07b).The Sloan Digital Sky Survey (henceforth SDSS; Myers et al.York et al. 2000) has renewed interest in binary quasars,and Hennawi et al. (2006; henceforth H06) have usedSDSS data to categorically confirm earlier evidence (e.g.,Djorgovski 1991; Hewett et al. 1998) that quasar clus-tering is enhanced on comoving scales ∼ < h − kpc(proper scales of ∼ < h − kpc at z ∼ . g ∼ <
21, redshifts of z ∼ < .
5, andcomoving separations ∼ < h − kpc (given the samplesize in M07b). Deep, wide imaging, such as from theSDSS, is thus key to testing predictions of the nature ofbinary quasars; for instance, by studying quasar cluster-ing as a function of redshift or luminosity. If enhancedsmall-scale quasar clustering is due to the enhanced biasof major galaxy mergers, rather than tidal forces exac-erbating quasar activity, then there should be no red-shift evolution in the relative bias of quasar clusteringon large and small scales. Further, some merger-drivenmodels predict stronger small-scale clustering for quasarsthan for lower-luminosity AGNs, due to different fuelingmechanisms (e.g., Hopkins et al. 2007b).Binary quasar samples are typically assembled frompairs that are initially targeted as possible gravitationallenses (e.g., Mortlock et al. 1999; H06). Because of this,the two members of most known binary quasars havesimilar colors. Although color similarity may be optimalin detecting lenses, scatter in the quasar color-redshiftrelation (e.g., Richards et al. 2001) dictates that strictcolor similarity cannot select all binary quasars. As bi-nary quasars are useful in testing merger-driven modelsof quasar activity, it is disconcerting that color similar-ity cuts might discard particularly informative binaries,such as any that are being exacerbated by tidal forces inmerging galaxies. Further, as binary quasars are scarce,relaxing color criterion might help provide enough bina-ries to study their redshift evolution.We have now performed spectroscopy of a sam-ple drawn from a complete, UVX ( u − g <
1) set of SDSS Data Release 4 (henceforth DR4;Adelman-McCarthy et al. 2006) binary quasar candi-dates (see Table 1 of M07b). The sample, ∼
45% of whichhas now been identified, is photometrically selected us-ing the Kernel Density Estimation (KDE) technique ofRichards et al. (2004). Our aim is to compile an exten-sive, homogeneous set of binary quasars, with proper sep-arations ∼ < h − kpc, mainly to study quasar cluster-ing on small scales, ultimately as a function of redshift.Most of our binary candidates are unlikely to be observedin the main SDSS quasar survey (e.g., Schneider et al.2007), because our observations probe fainter, and be-cause SDSS fibers (in single tiles) cannot be placed closerthan 55 ′′ . Our approach differs from previous stud-ies of binary quasars. In particular, our main goal isto study binary quasars, not gravitational lenses (un-like the samples compiled in, e.g., Kochanek et al. 1999;Mortlock et al. 1999; H06). Thus, excepting our initialUVX cut, our binaries are the first sample selected re-gardless of the relative colors of the component quasars.This allows an investigation of whether color similaritycan be used to optimally target binary quasars ( § §
2, we discuss our initial results in compiling a ho-mogeneous, spectroscopic binary quasar sample from theSDSS, and report 27 sets of new observations of binary quasar candidates. In § complete spectroscopic sample of bi-nary quasars. All our binaries are selected from the DR4KDE candidates of M07b, and are therefore a straightfor-ward subset of SDSS DR4, making our selection functionvery simple. We correct all magnitudes for Galactic ex-tinction using the maps of Schlegel, Finkbeiner & Davis(1998), unless otherwise noted. We adopt Ω m = 0 . Λ = 0 .
73, consistent with WMAP3 (Spergel et al.2007), and h ≡ H / − Mpc − = 0 .
7. We de-note transverse proper (comoving) scales as R prop ( R )and radial proper (comoving) scales as s prop ( s ). DATA
Observations
Candidate Selection
Our candidate binary quasars are photometric objectsin SDSS DR4 that are classified as quasars by the KDEtechnique (Richards et al. 2004), have u − g < g < ′′ to 6 ′′ of another such object. The SDSS ugriz filters are described in Fukugita et al. (1996). Asin M07b, we inspect our candidates and discard anythat are clearly not quasars (close KDE candidates areoccasionally misclassified H II regions in low-redshiftgalaxies). We restrict our spectroscopy, and analysis, toquasar pairs with angular separations of ∆ θ > ′′ , so thatcomponents of a pair appear clearly separated in SDSSimaging (e.g., H06). Although scales that are comparableto small galaxy groups or smaller are most useful in test-ing merger models of quasar activity, our upper limit of∆ θ < ′′ is somewhat arbitrary, providing a reasonablenumber of candidates for a small spectroscopic study.Table 1 of M07b lists our 98 DR4 candidate binaries (anda further 13 candidates with ∆ θ < ′′ ). At g <
21 theKDE technique, coupled with a UVX ( u − g <
1) cut,selects quasars with 95% efficiency, and is over 95% com-plete for redshifts of 0 . ∼ < z ∼ < . z ∼ < .
5, our selection should thus only bias our sam-ple against those binaries where one or both componentquasars is reddened beyond u − g = 1. Spectroscopy
Spectroscopy of our DR4 KDE binary quasar candi-dates was obtained with the R-C Spectrograph on theMayall 4-m, over 5 nights (UT 2007 February 22–26)at Kitt Peak National Observatory. We used a 1 . ′′ by98 ′′ long-slit set at the position angle of the candidate bi-nary, allowing both components to be simultaneously ob-served. The KPC-10A grating and WG360 blocking filteryielded a resolution of ∼ ∼ ∼ < . ′′ , allowing even our closest binary candidates( ∼ ′′ ) to be spatially separated. The survey goals werea positive identification and redshift for each DR4 KDEbinary candidate. This typically required a 15 minuteexposure when the faintest member of the candidate bi-nary was at g ∼ . g ∼
21, although the ∼ >
70% illuminatedMoon on February 25 and 26 typically prevented our g ∼
21 candidates from being spectroscopically identi-fied.Spectra were reduced at the telescope, using IRAF .Exposures ceased once a binary candidate could be iden-tified as; (1) containing one star or galaxy; or (2) con-taining two quasars with established redshifts. We esti-mate redshifts using the rest frame emission wavelengthslisted in Table 2 of Vanden Berk et al. (2001). Basedon deriving redshifts, where possible, from several differ-ent lines in each quasar’s spectrum, our typical error is∆ z ∼ . ∼
370 km s − , with little redshift de-pendence. We note that this is likely an overestimate, asvelocity differences between quasars are usually found tobe more precise when cross-correlating the full spectra,rather than measuring line shifts (e.g., Tonry & Davis1979; Djorgovski & Spinrad 1984).Tables 1a–1d detail our new observations. Table 1alists objects for which we obtained an identification foronly one member of the candidate binary. Table 1b listsconfirmed binary quasars. Following H06, we classifyquasar pairs with a line-of-sight velocity difference of | ∆ v k | < − ( s prop < . h − Mpc at ¯ z = 1 . § A g ∼ < .
17. Ofthese identified objects, 44 are both members of 22 candi-date quasar pairs, and 5 are objects from pairs for whichwe identified only one member. Of the 49 identified ob-jects 46 are quasars, confirming the KDE technique is ∼
95% efficient for A g ∼ < .
21 (Myers et al. 2007a). Ofthe 22 candidate quasar pairs for which we identified bothcomponents, 3 are quasar-non-quasar pairs, 9 are pro-jected quasar pairs (i.e. at disjoint redshifts), and 10 arebinary quasars. Several of the 10 binary quasars could,in fact, be previously unrecorded lenses; see § ′′ ≤ ∆ θ < ′′ ) DR4 KDE candidates havenow been spectroscopically identified (see Table 3). Withthe caveat that bright objects may have been observedfirst, ∼
42% of the candidate pairs are binary quasars,and only ∼
16% of the pairs contain a non-quasar.
Interesting Spectroscopic Pairs Distributed by the National Optical Astronomy Observatory,which is operated by the Association of Universities for Research inAstronomy, Inc., under cooperative agreement with the NationalScience Foundation.
Potential Lenses
Five pairs in Table 1b have sufficiently similar spec-tra, at our ∼ ′′ ≤ ∆ θ < ′′ are rare (e.g.,Inada et al. 2007), particularly for z ∼ <
2, we interpretthese objects as binaries (see also Kochanek et al. 1999),although they certainly merit higher-resolution spec-troscopy. A lensing interpretation is especially unlikelyfor the three possible lenses used in our clustering anal-ysis ( § ∼ ′′ separations and, in two cases,dissimilar colors. In Figure 1 we display the spectra ofour most likely lens candidates. SDSSJ1158+1235A andB, in particular, have almost identical spectra at our res-olution. Notes on Ambiguous Binaries
Table 1d, lists four candidates that we could not defini-tively identify. We conclude that two of these objectsare binaries, for the following reasons, quoting all wave-lengths in the observed frame.SDSSJ093424.32+421130.8 andSDSSJ093424.11+421135.0 consist of a quasar at z = 1 .
339 and a featureless spectrum (after 4200s ofexposure). SDSSJ093424.11+421135.0 is faint (observed g = 21 . g = 20 .
39) hasbroad emission at 4320˚A, and near 7840˚A at the rededge of our coverage. Although the brighter object( g = 20 . ∼ < II H and K emission. Assuming the proper motion ofthe M star(s) is moderate, this object may look more orless like an M star-quasar pair over time.SDSSJ1507+2903A,B have strong emission near5250˚A and 5215˚A, respectively. Neither spectrum hasadditional features over 3800–7500˚A, and so we assumethat the emission is Mg II , placing both quasars at z ∼ .
87. The ambiguity for this pair is that their redshiftsimply | ∆ v k | = 2100 km s − . As a shift of δz < . | ∆ v k | < − , we identify thispair as a binary. Myers et al. COLOR SELECTION OF BINARY QUASARS
To study relative color selection of binary quasars, weuse the χ color similarity statistic introduced by H06 χ color ( A ) = X ugriz ( f i − Af i ) [ σ i ] + A [ σ i ] (1)where the subscripts 1 and 2 represent the componentsof a pair. The superscript i refers to flux ( f ) in the 5SDSS bands ( ugriz ). For asinh magnitudes ( m ) f i = 2 F b i sinh (cid:2) − m i /P − ln b i (cid:3) σ if = ( σ im /P ) p (2 F b i ) + ( f i ) (2)where P = 2 . / ln 10 (Pogson 1856), F = 3630 . b [ u,g,r,i,z ] = [1 . , . , . , . , . × − (Lupton et al.1999; Stoughton et al. 2002). A quasar pair with moresimilar colors has a lower χ color . Iterative equations forcalculating A in Equation 1 can be ill-conditioned for χ color ∼ >
30, so, throughout this work, we numericallydetermine A by bisection.Binary quasars are often pairs rejected from grav-itational lens searches, and as such, the componentsof known binaries typically have very similar col-ors, a long-known example being SDSSJ1637+2636A,B( χ color = 2 .
8; see Table 2; Sramek & Weedman 1978;Djorgovski & Spinrad 1984). Schemes designed to opti-mize binary quasar searches by selecting pairs with simi-lar colors, will, therefore, naturally reselect known binaryquasars. Our DR4 KDE objects are simply all candidateswith a high probability of being quasars and thus, afterthe initial homogeneous UVX cut, are selected irrespec-tive of the relative colors of the components of the pair.The UVX cut itself, at z ∼ < .
5, should only bias oursample against those binaries with a component that isintrinsically dust-reddened beyond u − g = 1. Our sam-ple should thus be useful in determining color similaritycuts to optimize binary quasar selection. However, someof the quasar pairs in Table 2 were selected by H06 tohave χ color <
20; as we avoided reobserving these pairs,our data in Tables 1b–1d may be biased to χ color > χ color values of the DR4 KDE binary candi-dates identified to date fairly represent the full sam-ple. We compare the cumulative fraction of the 45identified candidates (i.e., Table 3) to the remaining 53(3 ′′ < ∆ θ < ′′ ) candidates, as a function of χ color . Atwo-sample Kolmogorov-Smirnov test cannot distinguishthe distributions, suggesting that the colors of the ob-served candidates fairly represent all candidates. Theupper-right panel of Figure 3 compares the χ color cu-mulative probability for the 21 confirmed binary quasars(or lenses) and the 24 confirmed non-binaries (projectedquasar pairs, star-quasar pairs, NELG-quasar pairs).The K-S test probability that these two distributions aredrawn from the same underlying χ color distribution is ∼ χ color can indeed discriminatebinary quasars from non-binaries.As our candidates are selected without a χ color cut, wecan ask what χ color limit optimizes completeness (num-ber of binaries/21 total binaries) and efficiency (num- ber of binaries/45 observed candidates). To sample ∼ >
50% ( ∼ > χ color ∼ < χ color ∼ < χ color ∼ < χ color <
70 cut rejects all but one quasar-starpair, while retaining all quasar-quasar projections.As there is reasonable scatter in the quasar color-redshift relation (e.g., Richards et al. 2001) full photo-metric redshift (henceforth photoz ) information shouldbetter select binary quasars. To test this, we consider theprimary “CZR” photoz solution (Weinstein et al. 2004;“ z phot range” in Tables 1–2) for each quasar in our sam-ple. We determine the overlap fraction of the primary photoz solutions of the two quasars in a pair, multiplyingby the probability that the quasar occupies that primarypeak.The completeness and efficiency of a binary quasarsample obtained by considering photoz overlap are plot-ted in the lower-left panel of Figure 3. Confirmed non-binaries typically have no photoz overlap. A proba-bility cut at >
3% overlap will return 90% of bina-ries and is 73% efficient. The two binaries that aremissed are the “ambiguous” SDSSJ1235+6836A,B, andSDSSJ1637+2636A,B, the “A” component of which hasa poorly behaved photoz solution. If one additionally ob-served all candidates that contained a quasar with a poor photoz (characterized by a probability of < . photoz overlap is a more ef-ficient mechanism for selecting binaries, it does so at theexpense of projected quasar pairs. A cut at <
3% over-lap could therefore be used to discard binary quasars infavor of projected pairs.In conclusion, if a survey’s goal is to select both binaryquasars and projected quasar pairs, a cut of χ color ∼ < ∼
95% efficient(Richards et al. 2004; Myers et al. 2006, 2007a). Cuts of χ color ∼ <
20, or stricter, may be necessary without priorKDE photometric classification but will miss ∼ >
33% ofall binary quasars. Interestingly, 2 ( ∼ > .
03 in the overlap of the two quasars’ photozs is superior to χ color selection. We stress that thisanalysis applies only to quasars pre-selected using an ef-ficient photometric classification technique. PROJECTED QUASAR CLUSTERING AT 25 h − kpc With our new observations (Table 1b and 1d), we havenow identified all DR4 KDE binary quasar candidateswith A g < . g < .
85 and 3 . ′′ < ∆ θ < . ′′ . InTable 4 we compile the binaries that meet these crite-ria. The three possible lenses in this subsample are wide-separation (∆ θ ≥ . ′′ ), making them likely binaries. Asour sample is complete over 3 . ′′ < ∆ θ < . ′′ , it is alsoeffectively complete for quasars in SDSS DR4 with sepa-rations on proper scales of 23 . < R prop < . h − kpcover 1 . < z < . DD/DR estima-tor (e.g., Shanks et al. 1983) for quasar-quasar ( QQ ) paircounts compared to expected quasar-random ( QR ) paircounts W p = QQ h QR i − h QR i in Equation 3 by constructinga catalog of random points with the same angular cov-erage as SDSS DR4, correcting the SDSS data for maskholes, as in Myers et al. (2006, 2007a). We further limitour random catalog to areas of the sky with Galactic ab-sorption A g < .
17. We assign random points a redshiftaccording to a fit to the normalized redshift distributionof ( A g < . g < .
85) quasars in the DR1 catalog(Schneider et al. 2003), from which the DR4 KDE quasarclassification is trained. Figure 7 of Myers et al. (2006)is similar to this redshift distribution, and Myers et al.(2006) argue that including additional quasars that over-lap the KDE color space minimally impacts this N ( z )distribution. To represent our fit, we use a modifiedGaussian d N = β exp − | z − ¯ z | n nσ ni d z (4)and find a best fit with ¯ z = 1 . σ = 0 . β = 0 .
65 and n = 3 (see also Myers et al. 2007a). We have repeatedour analyses instead using a spline fit, and our resultsdiffer by ∼ < h QR i in a given bin we use a random cat-alog with 1000 times as many points as the ( A g < . g < .
85) KDE DR4 photometric quasar catalog. Wetotal all QR counts in the angular bin of interest, andnormalize the result (divide by 1000). We then createa random catalog with points distributed according toEquation 4, and determine the fraction of pairs thatwould lie within | ∆ v k | < − by Monte Carlosampling to 0.1% precision. Multiplying this fraction bythe normalized QR counts yields h QR i . In a bin of 3 . ′′ < ∆ θ < . ′′ we expect 30.3 QR counts. Over our full N ( z ),we expect 0.0166 pairs with | ∆ v k | < − . Thus h QR i = 0 . QQ = 16 for the 8 (non-unique) pairs in Table 4. The implied projected corre-lation function averaged over 3 . ′′ < ∆ θ < . ′′ is thus W p = 30 . ± . . . We note that, for all candidate pairs, QQ (3 . ′′ < θ < . ′′ ) = 40, yielding an angular corre-lation of ω ( θ ) = (40 / . − . ± . . , consistentwith M07b.One of the pairs (SDSSJ1507+2903) in Table 4 has | ∆ v k | < − . We reasonably include this pair in our analysis, given the precision of our redshift esti-mates. Instead rejecting SDSSJ1507+2903 and assum-ing QQ = 14 would lower our estimate of W p by ∼ | ∆ v k | < − limit from H06 is intendedto bracket possible shifts in quasar lines, and thus in-corporate possible errors on | ∆ v k | but is otherwise arbi-trary. Although our Monte Carlo sampling of | ∆ v k | < − pairs does not model any error in | ∆ v k | ,relaxing this velocity window to | ∆ v k | < − would imply h QR i = 0 .
533 for QQ = 16, lowering our W p estimate by ∼ W p impliedby relaxing our velocity criterion is far smaller than ourerrors on W p , we proceed including SDSSJ1507+2903 inour analyses and maintaining | ∆ v k | < − inour Monte Carlo sampling.We apply our approach to different redshift rangesby weighting the number density of points in the ini-tial calculation of QR by the relative fraction of DR4KDE quasars obtained by integrating under Equation 4.We can thus determine h QR i over 1 . < z < . . < R prop < . h − kpc. In Figure 4 we comparethe small-scale clustering of quasars determined from ourcomplete DR4 binary quasar sample to the results fromH06. In the left-hand panel, we consider our calcula-tion for all quasars in Table 4, and project the result forboth proper and comoving scales by placing all binariesat the mean proper or comoving distance of our sample.In the right-hand panel, we consider binaries in Table 4with 23 . < R prop < . h − kpc and 1 . < z < . . < R prop < . . < z < . < h − kpc ( < h − kpc), andintegrate them over our scales of interest. At the meanredshift (¯ z = 1 .
40) of our sample, 3 . ′′ < θ < . ′′ , therange for which our sample is complete, is equivalent toproper (comoving) scales of 23 . h − kpc < R prop < . h − kpc (56 . h − kpc < R < . h − kpc). Overthis angular range, we find W p = 30 . ± . . for our data.The proper (comoving) power-law fit to the data fromH06, implies W p = 55 . . σ (1 . σ ) difference. The difference is slightly morepronounced if we determine W p for the H06 data at themean scale of the 8 binaries in our sample, instead of pro-jecting 3 . ′′ < θ < . ′′ back to z = 1 .
4. For the 5 bina-ries in our spatially complete clustering subsample, whichcovers scales of 23 . h − kpc < R prop < . h − kpc,we find W p = 24 . ± . . , and the H06 data implies W p = 57 .
2, 2.4 times (and 2 . σ ) higher than our result. DISCUSSION
We find that the projected correlation function ofquasars at proper scales of ∼ h − kpc has an am-plitude a factor of 2.4 times lower than that deter-mined by H06. H06 argue that clustering on these scalesis ∼
10 times higher than expected from projecting Myers et al.the quasar autocorrelation of Porciani et al. (2004) tosmaller scales. Our data thus imply excess quasar clus-tering at ∼ h − kpc of a factor of ∼
4, consistent withthe quoted upper limit in M07b. Given that our sampleis targeted differently than any previous samples of bi-nary quasars (i.e. UVX but otherwise regardless of thecolor similarity of the candidate components), and givenour simple selection function, our work might be viewedas independently corroborating the evidence for excessquasar clustering on small scales first detected in H06.We note that our binary quasar clustering subsample islargely independent of the sample used by H06, as of the8 binaries listed in Table 4 only 2 appear in Table 2.At R prop ∼ h − kpc we find a significantly smallerclustering amplitude than found by H06. As the bina-ries in Table 4 are all at z < R prop ∼ h − kpc binaries with z >
2, an in-teresting possibility is that W p is a function of scale and redshift. Certainly, on large scales ( ∼ > h − Mpc),quasars cluster twice as strongly at z > z < z = 1 .
87 (J. Hennawi2007, private communication) compared to ¯ z = 1 . ∼ > h − Mpc) scalesto follow b ( z ) = 0 .
53 + 0 . z ) . Given that quasarclustering scales as b , the implied relative amplitude be-tween quasar clustering in H06 and our sample is ∼ . W p = 30 . ± . . by this fac-tor implies W p = 53 . ± . . , easily consistent with thevalue of W p = 57 . W p = 24 . ± . . estimate for ourcomplete subsample (¯ z = 1 .
60) in the same way implies W p = 33 . ± . . , lower than W p = 57 . ∼ σ . Thus, multiplying our quasar clustering am-plitudes on small scales by the implied bias evolutionfrom quasar clustering on large scales somewhat recon-ciles our clustering amplitudes with the higher redshiftresults from H06. Consistent evolution of quasar clus-tering on large and small scales is a natural feature ofmerger-driven models of quasar origin (e.g., Figure 17 ofHopkins et al. 2007b).We can also look for direct redshift dependence withinour sample. Splitting the sample in Table 4 at z = 1 . W p = 27 . ± . . for z < .
55 (¯ z ∼ .
07) and W p = 34 . ± . . for z > .
55 (¯ z ∼ . z < .
55 ( z > .
55) is R prop = 23 . ∼ more clustered at z < .
55 than at z > .
55. Incorporating this ∼ W p at z > .
55 and z < .
55 to be 1 . ± .
54. Althoughnot a very significant detection, a factor of 1 . ± . b ( z ) = 0 .
53 + 0 . z ) bias relationimplies a ratio of ∼ . W p at z = 1 .
76 and z = 1 . ′′ ≤ ∆ θ < ′′ ) DR4 KDE binary quasar candidatesshould contain ∼
40 binary quasars. Based on our W p estimates, such a sample is the minimum necessary todetect any redshift dependence to quasar clustering at R prop ∼ h − kpc, providing only a ∼ . σ detec-tion. Completing this sample on a 4-meter class tele-scope would likely require dark time, as 14 of the DR4binary candidates have a component at g >
21. Alterna-tively, Shen et al. (2007) suggest a survey of quasar pairsat z >
3, where stronger quasar clustering may lead to amore significant detection of evolution. The DR6 KDEquasar catalog (Richards et al, in prep), which will beselected to i <
21 ( g ∼ < . ∼ . ∼ < z ∼ <
5) with which to pursue this goal in a singlehomogeneous sample. CONCLUSIONS
We have presented a spectroscopic survey of binaryquasar candidates with separations in the range 3 ′′ ≤ ∆ θ < ′′ . Our candidates (see Table 1 of M07b) are asubsample of quasars photometrically classified in SDSSDR4 using the KDE technique of Richards et al. (2004).We define a binary as a quasar pair with a line-of-sightvelocity difference of | ∆ v k | < − (see H06).We present 27 new sets of observations and identify bothmembers of 22 candidate binary quasars. Of the 22 newpairs, 10 turn out to be binary quasars (of which ∼ ′′ ≤ ∆ θ < ′′ at z ∼ < g <
21. A further 9 of our observed candi-dates are projected quasar pairs, and 3 contain a NELGor star. This confirms that the KDE technique is ∼ ∼
47% are binaries or lenses, ∼
40% are pro-jected quasar pairs, and the remainder contain a non-quasar.As our candidate binaries, beyond a UVX cut, are se-lected regardless of the relative colors of the quasars inthe pair, we can try to assess the color similarity criteriathat optimally select binary quasars. For quasars pre-selected with an already efficient approach such as theKDE technique, we find that a χ color similarity statis-tic of χ color <
70 will return 97% of binaries, lenses andprojected pairs (multiplied by the 95% completeness ofthe KDE technique itself) at 93% efficiency. Most of thisefficiency in selecting quasar pairs comes from the KDEtechnique itself, as imposing no color similarity criterionis 87% efficient. To select binary quasars, while reject-ing projected quasar pairs, we recommend a cut in theoverlap of the photometric redshifts of the two candidatequasars in a pair. An overlap of ∼ > .
03 in the primarysolution for the photometric redshift probability densityfunctions of the pair can be constructed to be ∼ ∼
70% efficient for binary quasars. Sim-ilarly, of course, the reverse probability cut of ∼ < . χ color <
70 cut to remove stars,can be used to reject binary quasars in favor of projectedquasar pairs.We measure the clustering of a complete sample of DR4binaries on proper scales of 23 . < R prop < . h − kpc.We find that, at ∼ h − kpc, quasars cluster with anamplitude 2.4 times, or 2 . σ , lower than determined byH06. As the mean redshift of the H06 sample is ¯ z = 1 . z = 1 .
40 for our sample, this can be inter-preted as evidence of evolution in quasar clustering onscales of ∼ h − kpc. The implied evolution is broadlyconsistent with merger-driven models, where the quasarpopulation is expected to evolve with consistent large-to-small scale clustering (e.g., Hopkins et al. 2007b). Wefind no significant evidence for quasar clustering evolu-tion at ∼ h − kpc, from ¯ z ∼ .
07 to ¯ z ∼ .
76, in oursample alone. Assuming evolution in the binary quasarpopulation at the level suggested by our current sample,we argue that the 98 quasars in the DR4 KDE candi-date binary sample will detect any clustering evolutionat proper scales of ∼ h − kpc at ∼ . σ significance.A sample of ∼
200 candidates ( ∼
80 binary quasars) willbe necessary to definitively detect clustering evolution at ∼ h − kpc for z ∼ < . REFERENCESAdelman-McCarthy, J. K., et al. 2006, ApJS, 162, 38Antonucci, R. 1993, ARA&A, 31, 473Bahcall, J. N., Bahcall, N. A., & Schneider, D. P. 1986, Nature,323, 515Bahcall, J. N., Kirhakos, S., Saxe, D. H., & Schneider, D. P. 1997,ApJ, 479, 642Barnes, J. E., & Hernquist, L. 1996, ApJ, 471, 115Cao, L., Wei, J.-Y., & Hu, J.-Y. 1999, A&AS, 135, 243Croom, S. M., & Shanks, T. 1996, MNRAS, 281, 893Croom, S. M., et al. 2005, MNRAS, 356, 415Djorgovski, S. 1991, in ASP Conf. Ser. 21, The Space Distributionof Quasars, ed. D. Crampton (San Francisco: ASP), 349Djorgovski, S., & Spinrad, H. 1984, ApJ, 282, L1Djorgovski, S. G., Courbin, F., Meylan, G., Sluse, D., Thompson,D., Mahabal, A., & Glikman, E. 2007, ApJ, 662, L1Elvis, M. 2000, ApJ, 545, 63Fukugita, M., Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K.,& Schneider, D. P. 1996, AJ, 111, 1748Gehrels, N. 1986, ApJ, 303, 336Hennawi, J. F., et al. 2006, AJ, 131, 1 (H06)Hewett, P. C., Foltz, C. B., Chaffee, F. H., Francis, P. J.,Weymann, R. J., Morris, S. L., Anderson, S. F., & MacAlpine,G. M. 1991, AJ, 101, 1121Hewett, P. C., Foltz, C. B., Harding, M. E., & Lewis, G. F. 1998,AJ, 115, 383Hopkins, P. F., & Hernquist, L. 2006, ApJS, 166, 1Hopkins, P. F, Hernquist, L., Cox, T. J., Di Matteo, T.,Robertson, B., & Springel V. 2006, ApJS, 163, 1Hopkins, P. F., Lidz, A., Hernquist, L., Coil, A. L., Myers, A. D.,Cox, T. J., & Spergel, D. N. 2007, ApJ, 662, 110Hopkins, P. F., Hernquist, L., Cox, T. J., & Keres, D. 2007,ArXiv e-prints, 706, arXiv:0706.1243Inada, N., et al. 2003, Nature, 426, 810Kochanek, C. S., Falco, E. E., & Mu˜noz, J. A. 1999, ApJ, 510, 590Landy, S. D., & Szalay, A. S. 1993, ApJ, 412, 64 Lupton, R., Gunn, J. E., & Szalay, A. S. 1999, AJ, 118, 1406Mason, K. O., et al. 2000, MNRAS, 311, 456Mortlock, D. J., Webster, R. L., & Francis, P. J. 1999 MNRAS,309, 836Myers, A. D., et al. 2006, ApJ, 638, 622Myers, A. D., et al. 2007a, ApJ, 658, 85Myers, A. D., et al. 2007b, ApJ, 658, 99 (M07b)Oguri, M. 2006, MNRAS, 367, 1241Oguri, M., et al. 2005, ApJ, 622, 106Inada, N., et al. 2007, ArXiv e-prints, 708, arXiv:0708.0828Peng, C. Y., et al. 1999, ApJ, 524, 572Phinney, E. S., & Blandford, R. D. 1986, Nature, 321, 569Pogson, N. R. 1856, MNRAS, 17, 12Porciani, C., Magliocchetti, M., & Norberg, P. 2004, MNRAS,355, 1010Richards, G. T., et al. 2001, AJ, 121, 2308Richards, G. T., et al. 2004, ApJS, 155, 257Rusin, D., 2002, ApJ, 572, 705Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500,525Schneider, D. P., et al. 2003, AJ, 126, 2579Schneider, D. P., et al. 2007, AJ, 134, 102Shanks, T., Bean, A. J., Ellis, R. S., Fong, R., Efstathiou, G., &Peterson, B. A. 1983, ApJ, 274, 529Shen, Y., et al. 2007, AJ, 133, 2222Spergel, D. N., et al. 2007, ApJS, 170, 377Sramek, R. A., & Weedman, D. W. 1978, ApJ, 221, 468Stoughton, C., et al. 2002, AJ, 123, 485Tonry, J., & Davis, M. 1979, AJ, 84, 1511Weinstein, M. A., et al. 2004, ApJS, 155, 243Vanden Berk, D. E., et al. 2001, AJ, 122, 549Veron-Cetty, M.-P., et al. 2004, A&A, 414, 487York, D. G., et al. 2000, AJ, 120, 1579
Myers et al.
Fig. 1.—
Two of the five quasar pairs in our DR4 KDE sample that require higher resolution spectroscopy to determine whether they arebinary quasars or a lensed quasar. These spectra were taken with the R-C Spectrograph on the Mayall 4-m at KPNO at a resolution ∼ θ < ′′ . In both panels, component A is the upper spectrum and componentB has been offset, which can cause component B to have a flux density below zero. In each case, both components are, in reality, at nearlyidentical flux levels. Fig. 2.—
Spectra of SDSSJ1235+6836A,B taken with the R-C Spectrograph on the Mayall 4-m at KPNO at a resolution ∼ component B offset by ± − , a typical window for a binary quasar(e.g., H06). The systemic redshift of component A in Table 1d is derived assuming the emission line near 3900˚A is C IV , and is close tothe red side of these windows ( z = 1 . Fig. 3.—
Upper-left:
The cumulative probability distribution of the χ color color similarity statistic (Equation 1) for the 45 spectroscopicallyobserved DR4 KDE candidate binary quasars, the 53 that are not yet observed, and for all 98 candidate pairs. Upper-right:
Similar toupper-left, but for the 21/45 observed candidates that are binary quasars (or lenses), for the 24/45 that are not binaries and for the 18/45quasar pairs at disjoint redshifts.
Lower-left:
The selection completeness and efficiency effects of imposing a χ color limit on the 45 observedDR4 KDE candidate binaries. Efficiency is (number of binaries < χ color )/(total candidates < χ color ); completeness is (number of binariesor lenses < χ color )/21. Lower-right:
Similar to lower-left, but using the fractional overlap of the photometric redshift solutions for thecandidate components ( z phot ) as the determinant of binary selection. Note that more overlap in photometric redshift implies greater colorsimilarity. Fig. 4.—
The projected correlation function, W p , of quasars on small scales. Left:
The results of H06 in proper (lower axis) and comoving(upper axis) coordinates are compared to our mean DR4 KDE spectroscopic results averaged over 3 . ′′ < ∆ θ < . ′′ and projected toa mean transverse separation at z = 1 . < h − kpc ( < h − kpc). Right:
A similar comparison but our data are now averaged over scalesof 23 . < R prop < . h − kpc and redshifts of 1 . < z < .
1. The DR4 KDE binary quasar sample is spectroscopically complete for A g < . g < .
85, and 3 . ′′ < ∆ θ < . ′′ , and for 23 . < R prop < . h − kpc over 1 . < z < . TABLE 1aDR4 KDE candidate binaries for which we have observed one member ∆ θ χ color Name α (J2000) δ (J2000) g z phot range SDSS z Our z5.50 36.2 SDSSJ093014.81+420038.7 09 30 14.816 +42 00 38.71 19.71 1.15,1.425,1.50,0.93 0.544 ? SDSSJ093015.01+420033.6 09 30 15.016 +42 00 33.68 20.03 1.85,2.025,2.15,0.625.55 24.2 SDSSJ093521.02+641219.8 09 35 21.020 +64 12 19.89 20.99 1.45,1.775,2.10,0.92 1.566SDSSJ093521.80+641221.9 09 35 21.807 +64 12 22.00 20.96 0.25,0.375,0.45,0.665.33 6.4 SDSSJ095840.74+332216.3 09 58 40.746 +33 22 16.31 19.18 1.35,1.475,2.10,0.84 1.891 1.888SDSSJ095840.94+332211.5 09 58 40.945 +33 22 11.59 20.64 1.45,1.725,2.10,0.913.46 161 SDSSJ103939.31+100253.0 10 39 39.317 +10 02 53.01 18.42 0.10,0.175,0.25,0.56 0.161 0.161SDSSJ103939.53+100254.3 10 39 39.532 +10 02 54.40 19.60 0.65,1.125,1.55,0.924.22 8.2 SDSSJ162847.75+413045.4 16 28 47.752 +41 30 45.45 19.81 1.35,1.525,1.70,0.85SDSSJ162848.06+413043.1 16 28 48.069 +41 30 43.19 20.40 1.95,2.075,2.20,0.63 0.831 ? Note . — The angular pair separations are denoted ∆ θ ( ′′ ), and χ color is each pair’s color similarity statistic (Equation 1). Thephotometric redshifts z phot are expressed as “lowest extent, peak, highest extent, probability of true redshift lying in this range”. The“SDSS z ” column shows matches to any spectroscopic object in the SDSS DR6 Catalog Archive Server (mainly, e.g., Schneider et al.2007). The “Our z” column lists our new spectroscopic confirmations from KPNO data. In the “SDSS z” and “Our z” columns theobject is a quasar at the provided redshift, unless otherwise noted. Redshifts labeled with a ? are based on a single emission line,which is reasonably assumed to be Mg II . g is not corrected for Galactic extinction. TABLE 1bConfirmed binary quasars in the DR4 KDE candidate sample ∆ θ χ color Name α (J2000) δ (J2000) g z phot range SDSS z Our z*3.56 4.0 SDSSJ1158+1235A 11 58 22.776 +12 35 18.59 19.90 0.45,0.525,0.65,0.55 0.596 ? SDSSJ1158+1235B 11 58 22.989 +12 35 20.31 20.12 0.40,0.475,0.65,0.63 0.596 ? *4.74 32.0 SDSSJ1320+3056A 13 20 22.545 +30 56 22.87 18.60 1.35,1.475,1.65,0.90 1.597 1.595SDSSJ1320+3056B 13 20 22.643 +30 56 18.29 19.92 1.45,1.575,1.80,0.51 1.596*4.50 2.8 SDSSJ1418+2441A 14 18 55.418 +24 41 08.92 19.27 0.45,0.525,0.70,0.96 0.573 0.572SDSSJ1418+2441B 14 18 55.536 +24 41 04.71 20.22 0.40,0.625,0.70,0.86 0.5734.27 3.2 SDSSJ1426+0719A 14 26 04.266 +07 19 25.86 20.82 0.95,1.175,1.45,0.99 1.312SDSSJ1426+0719B 14 26 04.326 +07 19 30.04 20.12 1.00,1.225,1.45,0.97 1.3095.41 25.6 SDSSJ1430+0714A 14 30 02.664 +07 14 15.62 20.27 1.00,1.225,1.40,0.97 1.246SDSSJ1430+0714B 14 30 02.886 +07 14 11.33 19.50 1.05,1.375,1.45,0.97 1.258 1.2615.14 65.6 SDSSJ1458+5448A 14 58 26.165 +54 48 14.85 20.79 1.50,1.775,1.95,0.75 1.913SDSSJ1458+5448B 14 58 26.728 +54 48 13.19 20.53 1.65,1.925,1.95,0.47 1.912*3.45 10.8 SDSSJ1606+2900A 16 06 02.812 +29 00 48.79 18.50 0.50,0.725,1.00,0.65 0.770 0.769?SDSSJ1606+2900B 16 06 03.021 +29 00 50.88 18.42 0.70,0.875,1.00,0.92 0.769?*4.92 25.6 SDSSJ1635+2911A 16 35 10.148 +29 11 20.65 18.83 1.45,1.575,1.80,0.84 1.586 1.582SDSSfJ1635+2911B 16 35 10.306 +29 11 16.19 20.43 1.40,1.525,1.85,0.79 1.590 Note . — We define a binary quasar by a line-of-sight velocity difference of | ∆ v k | < − in the rest-frame ofeither component. The pair containing SDSSJ143002.66+071415.6 has velocity difference | ∆ v k | = 2000 ±
400 km s − , justinside our definition of a binary. Components are denoted A and B so that the position angle from A to B lies between 0 ◦ and 180 ◦ . A preceding * denotes that our spectroscopy alone is insufficient to rule out a lens interpretation for this pair (see § g is not correctedfor Galactic extinction. See Table 1a for additional notes describing shared notation . TABLE 1cConfirmed projected pairs in the DR4 KDE candidate sample ∆ θ χ color Name α (J2000) δ (J2000) g z phot range SDSS z Our z4.28 1220 SDSSJ083258.34+323003.3 08 32 58.348 +32 30 03.35 19.96 2.75,2.775,2.80,0.88 0.397SDSSJ083258.56+323000.0 08 32 58.567 +32 30 00.08 19.59 0.40,0.425,0.50,0.70 star3.42 2.0 SDSSJ084257.37+473342.5 08 42 57.378 +47 33 42.56 19.00 0.50,0.625,0.70,0.51 1.552 1.554SDSSJ084257.63+473344.7 08 42 57.638 +47 33 44.74 20.45 0.85,1.775,2.10,0.84 1.6814.28 4.9 SDSSJ085914.77+424123.6 08 59 14.771 +42 41 23.67 21.02 0.95,1.375,1.65,0.66 1.396SDSSJ085915.15+424123.5 08 59 15.159 +42 41 23.58 19.22 0.80,0.975,1.40,0.93 0.898 0.902 ? ? ? SDSSJ123122.37+493430.7 12 31 22.378 +49 34 30.75 19.94 1.55,1.775,2.15,0.95 1.811 ? ? SDSSJ150657.18+505607.9 15 06 57.183 +50 56 07.92 19.75 2.00,2.225,2.40,0.45 2.204
Note . — SDSSJ112556.32+143148.0, the NELG, has redshift z=0.246. Subsequent to our observations, SDSSJ094309.36+103401.3appeared in Inada et al. (2007), with z = 1 . g is not corrected for Galactic extinction. The redshift for SDSSJ123122.37+493430.7is based on a single C IV emission line, with weak confirming C III ]. See Table 1a for additional notes describing shared notation . TABLE 1dAmbiguous pairs in the DR4 KDE binary quasar candidate sample ∆ θ χ color Name α (J2000) δ (J2000) g z phot range SDSS z Our z4.80 42.5 SDSSJ093424.11+421135.0 09 34 24.112 +42 11 35.05 21.01 1.45,1.675,2.20,0.90 f/lessSDSSJ093424.32+421130.8 09 34 24.324 +42 11 30.87 20.30 1.00,1.125,1.40,0.99 1.3393.95 10.7 SDSSJ120727.09+140817.1 12 07 27.100 +14 08 17.18 20.39 1.60,1.775,2.00,0.89 1.801 ? SDSSJ120727.25+140820.3 12 07 27.259 +14 08 20.38 20.27 1.55,1.775,1.95,0.86 1.8 ?? ?? SDSSJ1235+6836B 12 35 55.270 +68 36 27.07 19.70 0.50,0.625,1.10,0.51 1.5144.35 14.4 SDSSJ1507+2903A 15 07 46.909 +29 03 34.15 20.44 0.80,0.975,1.25,0.77 0.875 ? SDSSJ1507+2903B 15 07 47.234 +29 03 33.28 19.97 0.70,0.775,0.95,0.66 0.862 ? Note . — Redshifts marked ?? are derived from a single emission line. This differs from the ? notation as the redshift is basedon similar emission in the other component (rather than simply assuming that the emission is Mg II ). The ambiguities, and whywe conclude that SDSSJ1235+6836 and SDSSJ1507+2903 are binaries but the other pairs are not, are discussed in § g is notcorrected for Galactic extinction. See Table 1a for additional notes describing shared notation . TABLE 2Previously identified DR4 KDE binary quasar candidates ( ′′ ≤ ∆ θ < ′′ ) ∆ θ χ color Name α (J2000) δ (J2000) g z phot range zProjected pairs4.90 36.7 SDSSJ024907.77+003917.1 02 49 07.778 +00 39 17.12 19.36 2.00,2.175,2.25,0.48 2.164SDSSJ024907.86+003912.4 02 49 07.866 +00 39 12.40 20.63 0.45,0.675,0.85,0.84 star4.09 18.0 SDSSJ083649.45+484150.0 08 36 49.456 +48 41 50.08 19.31 0.45,0.675,0.80,0.67 0.657SDSSJ083649.55+484154.0 08 36 49.554 +48 41 54.06 18.50 1.50,1.675,1.95,0.94 1.7125.42 4.5 SDSSJ090235.35+563751.8 09 02 35.356 +56 37 51.84 20.95 1.15,1.275,1.45,0.98 1.39SDSSJ090235.73+563756.2 09 02 35.731 +56 37 56.29 20.56 1.05,1.225,1.45,0.98 1.343.14 37.1 SDSSJ095454.73+373419.7 09 54 54.735 +37 34 19.79 19.57 0.95,1.475,1.65,0.90 1.544SDSSJ095454.99+373419.9 09 54 54.999 +37 34 19.99 18.91 1.45,1.575,1.95,0.94 1.8924.76 10.2 SDSSJ114718.44+123439.8 11 47 18.448 +12 34 39.84 20.91 1.45,1.625,2.00,0.66 1.583SDSSJ114718.66+123436.3 11 47 18.668 +12 34 36.33 19.80 2.15,2.225,2.60,0.54 2.2323.06 5.6 SDSSJ120450.54+442835.8 12 04 50.543 +44 28 35.89 19.04 0.95,1.125,1.45,0.98 1.144SDSSJ120450.78+442834.2 12 04 50.784 +44 28 34.25 19.48 1.35,1.725,1.95,0.78 1.8145.04 23.3 SDSSJ124948.12+060709.0 12 49 48.127 +06 07 09.04 20.41 2.20,2.325,2.65,0.89 2.001SDSSJ124948.17+060714.0 12 49 48.179 +06 07 14.02 20.38 1.85,2.075,2.20,0.58 2.3763.01 20.0 SDSSJ125530.44+630900.5 12 55 30.445 +63 09 00.51 20.30 1.50,1.675,1.90,0.88 1.753SDSSJ125530.82+630902.0 12 55 30.823 +63 09 02.09 20.60 1.10,1.375,1.50,0.98 1.3934.94 31.2 SDSSJ142359.48+545250.8 14 23 59.484 +54 52 50.83 18.63 1.00,1.175,1.45,0.973 1.409 I SDSSJ142400.00+545248.7 14 24 00.006 +54 52 48.79 19.93 1.45,1.575,1.90,0.772 0.610 I I SDSSJ171335.03+553047.9 17 13 35.037 +55 30 47.91 19.11 2.00,2.175,2.20,0.686 star I S Binary quasars3.94 18.8 SDSSJ0959+5449A 09 59 07.060 +54 49 08.09 20.60 1.90,2.025,2.15,0.59 1.956SDSSJ0959+5449B 09 59 07.471 +54 49 06.38 20.07 1.40,1.575,2.10,0.90 1.9543.55 6.5 SDSSJ1259+1241A 12 59 55.464 +12 41 51.06 19.99 1.95,2.175,2.30,0.43 2.180SDSSJ1259+1241B 12 59 55.617 +12 41 53.81 20.09 1.90,2.175,2.25,0.52 2.1893.81 4.3 SDSSJ1303+5100A 13 03 26.144 +51 00 51.00 20.54 1.50,2.075,2.20,0.82 1.686SDSSJ1303+5100B 13 03 26.177 +51 00 47.21 20.37 1.60,1.775,2.00,0.93 1.6843.12 0.5 SDSSJ1337+6012A 13 37 13.085 +60 12 09.70 20.04 1.30,1.775,2.05,0.66 1.721SDSSJ1337+6012B 13 37 13.133 +60 12 06.60 18.59 1.50,1.625,1.95,0.94 1.7275.13 4.1 SDSSJ1432-0106A 14 32 28.949 -01 06 13.55 21.10 1.55,2.125,2.25,0.69 2.082SDSSJ1432-0106B 14 32 29.247 -01 06 16.06 17.83 1.90,2.025,2.15,0.96 2.0824.11 7.2 SDSSJ1530+5304A 15 30 38.564 +53 04 04.03 20.56 1.45,1.575,1.95,0.65 1.531SDSSJ1530+5304B 15 30 38.824 +53 04 00.65 20.70 1.40,1.725,2.15,0.93 1.5333.90 2.8 SDSSJ1637+2636A 16 37 00.881 +26 36 13.71 20.61 0.45,0.575,0.85,0.46 1.961 D SDSSJ1637+2636B 16 37 00.932 +26 36 09.87 19.36 1.40,1.525,1.80,0.64 1.961 D i SDSSJ1004+4112B 10 04 34.917 +41 12 42.81 19.04 1.55,1.725,2.15,0.79 1.734 i o SDSSJ1206+4332B 12 06 29.652 +43 32 20.61 19.38 1.95,2.175,2.35,0.53 1.789 o Note . — Components of a binary are denoted A and B so that the position angle from A to B lies between 0 ◦ and 180 ◦ .This convention differs from H06, from which we take identifications and redshifts, except for objects labeled S (taken fromthe SDSS), D (discovered by Sramek & Weedman 1978, confirmed as a possible lens by Djorgovski & Spinrad 1984, and likelya binary instead, e.g., Kochanek et al. 1999; Peng et al. 1999; Rusin 2002), i (part of the quad lens from Inada et al. 2003),o (Oguri et al. 2005), and I (Inada et al. 2007). Both quasars SDSSJ162902.59+372430.8 and SDSSJ162902.63+372435.1first appear in Mason et al. (2000). SDSSJ1004+4112A was discovered by Cao et al. (1999), and SDSSJ1432-0106B byHewett et al. (1991). We note that we mistakenly listed SDSSJ095454.73+373419.7 as lying at z=1.554 in M07b. g is notcorrected for Galactic extinction. See Table 1a for additional notes describing shared notation . TABLE 3Breakdown of ( ′′ ≤ ∆ θ < ′′ ) DR4 KDE quasar pairs Category Number of Confirmed PairsTotal binary quasar candidates 98Total now identified 45Likely binary quasars 19Quasar pairs separated in redshift 18Pairs containing ≥ Note . — It is possible that a few more objects listed as “Likelybinary quasars” may turn out to be a lensed quasar when scrutinizedat higher resolution.
TABLE 4Complete, statistical, clustering subsample
Name ∆ θ ( ′′ ) χ color R prop R z A z B | ∆ v k | TableSDSSJ0959+5449 3.94 18.8 23.8 70.2 1.956 1.954 200 (2)SDSSJ1320+3056 *4.74 32.0 28.8 74.7 1.595 1.597 200 (1b)SDSSJ1418+2441 *4.50 2.8 20.9 32.8 0.572 0.573 100 (1b)SDSSJ1426+0719 4.27 3.2 25.6 59.5 1.312 1.309 400 (1b)SDSSJ1458+5448 5.14 65.6 31.1 90.2 1.913 1.912 0 (1b)SDSSJ1507+2903 4.35 14.4 23.8 44.3 0.875? 0.862? 2100 (1d)SDSSJ1530+5304 4.11 7.2 24.9 63.1 1.531 1.533 200 (2)SDSSJ1635+2911 *4.92 25.6 29.9 77.3 1.582 1.590 900 (1b)
Note . — The DR4 KDE binary quasar candidate sample is now spectroscopically completefor component separations 3 . ′′ < ∆ θ < . ′′ for g < .
85 in regions with Galactic absorp-tion A g < .
17. 20 pairs meet these criteria, and 8 of them are (the listed) binary quasars.A * denotes a possible lens (see note in Table 1b). R prop ( R ) is the transverse proper (co-moving) separation ( h − kpc). | ∆ v k | is the line-of-sight velocity difference (km s − ). Thefinal column “Table” denotes where we first listed these binaries. The 5 listed binaries withtransverse separations of 23 . ≤ R prop ≤ . . < z < ..